This shows you the differences between two versions of the page.
Previous revisionNext revision | |||
— | university:courses:electronics:efficiency_power_loss [10 May 2019 20:17] – Add slide deck Mark Thoren | ||
---|---|---|---|
Line 1: | Line 1: | ||
+ | ====== Activity: Efficiency, Power Loss, and Thermal Management ====== | ||
+ | |||
+ | ===== Objective: ===== | ||
+ | |||
+ | The objective of this activity is to explore the concepts of efficiency, power loss, temperature rise, and heat flow. | ||
+ | |||
+ | ===== Safety: ===== | ||
+ | |||
+ | This experiment deals with power electronics, | ||
+ | |||
+ | ===== Background: ===== | ||
+ | All circuits require power. A smart phone is a close-to-home example; it is compact, reliable (hopefully) and performs an astonishing array of functions. Electric cars, network server blades, avionics, your laptop computer, and your microwave oven, all require power. And a common theme with most modern electronics is that powering them is not getting any easier - requirements on voltage accuracy, current capacity, size, cost, are all being pushed further and further. But no power conversion circuit is perfect, some power will be lost in the process, and it is this lost power, and what do do with it, that will be explored in this lab. | ||
+ | |||
+ | When thinking about power supplies, the term " | ||
+ | |||
+ | {{ : | ||
+ | <WRAP centeralign> | ||
+ | |||
+ | Here on Earth, there is the luxury of air to carry heat away from stuff that gets hot. What about a space application? | ||
+ | |||
+ | Whether the amount of power lost from a circuit is large or small, one thing is true - it WILL escape to the environment. It will escape by increasing the temperature of the circuit until the heat flowing out of the circuit is equal to the electrical power being lost, and the goal is to make sure that when this equilibrium is met, the circuit is still functioning properly. | ||
+ | |||
+ | With that in mind, let's do some experiments, | ||
+ | |||
+ | ===== Materials ===== | ||
+ | ADALM2000 Active Learning Module OR\\ | ||
+ | 2 multimeters (minimum), preferably with a 1A current range\\ | ||
+ | Solder-less breadboard\\ | ||
+ | Jumper wires\\ | ||
+ | PC/Mac running LTspice and Scopy\\ | ||
+ | 0-24V, 1Amp, Adjustable Power supply\\ | ||
+ | LT3080 LDO regulator\\ | ||
+ | LTM8067 Isolated Switching Regulator (on BOB)\\ | ||
+ | 6.2Ω, 10W power resistor\\ | ||
+ | TO-220 heat sink, Aavid 7021 or similar, or various sizes of double-sided, | ||
+ | Heat sink compound / thermal grease\\ | ||
+ | AD592 Temperature Sensor\\ | ||
+ | Optional: Infrared thermometer\\ | ||
+ | |||
+ | |||
+ | ===== Thermal Resistance Primer ===== | ||
+ | Why don't Linear Regulators have an efficiency number proudly displayed on the front page of the datasheet, like switching regulators? It could be because the relevant laws of physics are intuitively understood by most engineers using these parts - Current through any "black box", multiplied by the voltage drop, equals power that will leave the box somehow. In the case of an LDO regulator, that power leaves as heat. (If the "black box" were an LED, some of that power would leave as light, if it were a motor, the power might leave as mechanical power through the rotating shaft.) And if the input supply to an LDO regulator varies widely, the efficiency will also vary widely - it could be near 100% when the input supply is just a little bit higher than the output voltage, or 10% or less, if the input is 12V and the output is 1.2V. But there are definitely situations where linear regulators are the right tool for the job. (We'll save that discussion for later.) | ||
+ | |||
+ | Before even starting to build any circuitry, we know that we're going to have to get rid of some heat. The LT3080 regulator from the parts kit is in the very common T0-220 package, with a tab for mounting to a heat sink. | ||
+ | |||
+ | {{ : | ||
+ | <WRAP centeralign> | ||
+ | |||
+ | This shows the physical layout of the part, pinout, and three parameters, defined as follows: | ||
+ | |||
+ | **T< | ||
+ | |||
+ | **Θ< | ||
+ | |||
+ | **Θ< | ||
+ | |||
+ | Further defining terms: | ||
+ | |||
+ | **Thermal Resistance** - Resistance to the flow of heat, expressed as the temperature rise due to a given power flowing through the resistance. | ||
+ | |||
+ | **T< | ||
+ | |||
+ | **T< | ||
+ | |||
+ | **T< | ||
+ | |||
+ | These seemingly simple terms are in reality quite difficult to measure. Measuring " | ||
+ | |||
+ | This description from Vishay Application Note 827 illustrates this point: "For the MOSFET/heat sink assembly, a specially designed heat sink assembly of a copper block (4 in. x 4 in. x 0.75 in.) was used to simulate an infinite heat sink attached to the case of the TO-220 device." | ||
+ | |||
+ | Junction temperature is, as the name suggests, the temperature of the operational semiconductor junction in the device, which in reality may be many junctions in a complex circuit. And it is this temperature that must be kept below the maximum specified; if exceeded, the part is not guaranteed to function properly. But note that unless your device has a built-in temperature sensor (and some do), it is difficult to measure the junction temperature directly. | ||
+ | |||
+ | Note that the maximum junction temperature can be well above the boiling point of water - too hot to touch. So using your finger to test if a circuit is cool enough is not only dangerous, it is completely inaccurate. | ||
+ | |||
+ | So how are these numbers used? The objective is to keep the junction below the maximum allowed. So we can use knowledge of how much power is dissipated in the part (near the junction), and the thermal resistance to the air, to calculate how hot the junction will get. | ||
+ | |||
+ | T< | ||
+ | |||
+ | Where P< | ||
+ | |||
+ | One very useful mental model is to think of thermal resistances as electrical resistances, | ||
+ | |||
+ | 1°C/W = 1 Ω | ||
+ | |||
+ | 1W of dissipation = 1A of current being driven through the resistance | ||
+ | |||
+ | 1V = 1°C temperature rise across the resistance. | ||
+ | |||
+ | |||
+ | Doing a quick calculation on the LT3080 in the TO-20 package, if the input voltage is 10V, output voltage is 5V, and the load current is 200mA, the power dissipated in the part is (10V - 5V) * 0.2A = 1W. This will cause a temperature rise of 40°C, so if the air in the room is 25C, the junction will heat up to approximately 65°C - well under the 125°C maximum (but hot enough to burn skin!) The electrical analogy is shown below, running a DC operating point simulation. | ||
+ | |||
+ | |||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Notice that Tjunction is 65 " | ||
+ | |||
+ | But what happens if the load current increases to 500mA? Now you have to get rid of 2.5W, which will cause a temperature rise of 100°C, pushing you right up to the maximum junction of 125°C, with no safety margin. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | That doesn' | ||
+ | |||
+ | Θ< | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | And note that while Θ< | ||
+ | |||
+ | The LT3080 in the parts kit is the TO-220 package version, and we're not soldering it down, which means that we really ought to be using a heat sink. How does that affect our calculations? | ||
+ | |||
+ | *(This is not really luck, it's an essential piece of data.) | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | This shows the following: | ||
+ | |||
+ | Θ< | ||
+ | |||
+ | Θ< | ||
+ | |||
+ | This is reconcilable with table 5 above - the heat sink is a folded up piece of aluminum, with a total area of about 3060mm< | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Let's re-run the LTspice simulation one more time, with the Aavid 7021 heat sink's thermal resistance: | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | The expected temperature rise is about 32.5°C, for a junction temperature of 57.5°C | ||
+ | |||
+ | ===== Procedure: LT3080 Linear Regulator ===== | ||
+ | Refer to the circuit shown below. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | The LTspice file is set up to sweep the input voltage from 5V to 12V and plot input power, output power, and efficiency. Results are shown in Figure 10 below, with the red trace representing efficiency. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | As expected, efficiency is relatively high (about 66%) when the input voltage (shown in green) is low. As the input voltage increases, the power dissipation in the LT3080 (blue) increases, and efficiency decreases (to about 28% when the input voltage is 12V). Results of this simulation will reflect reality very accurately. The reason is that the loss mechanisms are straightforward - power dissipations are simply DC currents multiplied by DC voltages. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Construct the circuit on a solder-less breadboard, keeping the following in mind: | ||
+ | |||
+ | Mount the LT3080 to the heat sink first, with a small drop of heat sink compound between the package and heat sink. | ||
+ | Carefully twist the LT3080' | ||
+ | Note that the SET resistor is three 1M resistors in parallel. | ||
+ | WARNING: if the SET resistors lose contact, the output voltage will increase to its maximum, and the 6.2Ω resistor will get very hot! | ||
+ | |||
+ | Also, there are several options for measuring voltages and currents. Input voltage and current can be measured with multimeters set to appropriate voltage and current ranges, or, can be read directly from the power supply if it includes an accurate voltmeter and current meter. Output current can either be measured directly with a multimeter, or calculated, by first measuring the actual resistance of the load resistor with a multimeter and dividing the measured output voltage by the resistance. (The resistor in the parts kit has a 10% tolerance, so it should be measured first.) Input and output voltages can also be measured with the ADALM2000 and Scopy running in voltmeter mode, or with a multimeter. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Things are about to get a bit warm - too warm to touch. So we need a way of at least getting some idea of HOW warm without getting burned. The AD592 temperature sensor provides an easy way to do this: | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | The AD592 leads can be extended, and the middle lead is not connected so it can be used to provide extra support. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | A small rubber band can then be used to hold the sensor against the top surface of the LT3080 as shown in Figure 15. Use a tiny drop of thermal grease between the sensor and top of the LT3080 package. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | It was mentioned above that the top of the " | ||
+ | |||
+ | T< | ||
+ | |||
+ | With typical k values of 1.18 for DDPAK package (similar to TO-220) So while we don't have actual measurements for the LT3080, we can assume that the temperature rise of the die is about 20% greater than the temperature at the top package surface, measured with an AD592, small thermocouple, | ||
+ | |||
+ | Power up the circuit and fill out the following data table: | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | Note the relationship between LT3080 power dissipation, | ||
+ | |||
+ | ===== Procedure: LTM8067 Isolated Flyback DC/DC Converter ===== | ||
+ | Next, we'll explore the efficiency and thermal performance of the LTM8067 Isolated flyback module. We're not interested in the fact that it's isolated (meaning, output and input ground terminals are independent) or that it is a module (all components encased in a single package). We are interested in the fact that it is a switching converter, which is more efficient (and loses less power to the environment) than a linear regulator, at least under most circumstances. The LTM8067 in the parts kit comes mounted to a breakout board, with a potentiometer that allows the output voltage to be adjusted from 3V to 15V. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | The block diagram from the datasheet shows a basic isolated flyback circuit. Without going into details, one key point is worth worth noting: unlike the pass transistor in the LT3080, The MOSFET in the LTM8067 is either off completely, or on completely, operating as a switch. This means that very little power is dissipated in the transistor. Furthermore, | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Setup for this experiment is straightforward; | ||
+ | |||
+ | * Input positive is at the top left corner | ||
+ | * Input ground is at the lower left corner | ||
+ | * Output positive is bottom center | ||
+ | * Output negative is top center. | ||
+ | |||
+ | Also note that the output current capability of the LTM8067 varies with input voltage as shown below: | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Even with the BOB set to the minimum output of 3V, the 6.2 Ω power resistor will draw 440mA, requiring about 20V input voltage. Borrow a neighbor' | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Simulations of switching regulators are not as straightforward. Some aspects of the circuit' | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | However, we can still try extracting efficiency from the LTspice simulation. Disable the startup transient SPICE directives (right-click, | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | (Comparing with datasheet figure " | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | |||
+ | Construct the circuit on a solder-less breadboard. As with the LT3080 circuit, construction details matter. | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | <WRAP centeralign> | ||
+ | |||
+ | Power up the circuit and fill out the following data table: | ||
+ | |||
+ | {{ : | ||
+ | |||
+ | Note the relationship between LTM8067 power dissipation, | ||
+ | |||
+ | ===== Questions: ===== | ||
+ | How does the LTM8067 efficiency, power loss, and temperature rise compare to the LT3080? | ||
+ | |||
+ | The temperature rise vs. heat dissipated curve for the Aavid heat sink is slightly curved - it appears to have a lower thermal resistance as more heat is dissipated. Why? | ||
+ | |||
+ | <WRAP round download> | ||
+ | **Resources: | ||
+ | * Fritzing files: [[ https:// | ||
+ | * LTSpice files: [[ https:// | ||
+ | </ | ||
+ | |||
+ | =====Further Reading===== | ||
+ | Using .meas and .step commands to calculate efficiency in LTspice: [[https:// | ||
+ | |||
+ | ===== Slide Deck ===== | ||
+ | A slide deck is provided as a companion to this exercise, and can be used to help in presenting this material in classroom, lab setting, or in hands-on workshops. | ||
+ | <WRAP round download> | ||
+ | {{ : | ||
+ | </ | ||
+ | |||
+ | **Return to Lab Activity [[university: | ||